The most prestigious law school admissions discussion board in the world.
|
 |
 
|
so AI isnt really solving all of these problems, just stealing user input info?
Poast new message in this thread
Date: September 14th, 2026 9:58 AM Author: walnut rigor yarmulke (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134761) |
Date: September 14th, 2026 10:00 AM Author: Concupiscible fuchsia library so it's basically google? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134764) |
 |
Date: September 14th, 2026 10:10 AM Author: fragrant self-centered property (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134782) |
|
Date: September 14th, 2026 4:48 PM Author: lascivious sweet tailpipe *looks up the key companies developing AI*
*notices Google*
*nods* |
|
Date: September 14th, 2026 4:58 PM Author: Canary Hall yes, its google but friendlier, faster, and 20% more inaccurate. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135623) |
|
Date: September 14th, 2026 10:30 AM Author: aqua halford dilemma Yeah because it definitely could have solved a millennium problem just by looking at peoples prompts, which is why it took millions of tokens and a powerful internal model and why none of these users came forward with their own solutions! (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134814) |
|
Date: September 14th, 2026 10:41 AM Author: walnut rigor yarmulke https://www.washingtonpost.com/technology/2026/09/13/he-was-close-1-million-breakthrough-then-openai-swooped/ (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134838) |
|
Date: September 14th, 2026 10:48 AM Author: aqua halford dilemma You are so gullible if you believe the "perhaps the machine learned his secret work and used it against him" narrative. Open AI has already done an more extensive investigation and demonstrated this was impossible (article even refers to this). There was overlap by being in the same research program, but the final proof released by open ai was radically different and departed radically from what the guy was doing. Also lol at the idea he would have solved it. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134846) |
|
Date: September 14th, 2026 10:48 AM Author: walnut rigor yarmulke sup sam (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134849) |
|
Date: September 14th, 2026 11:52 AM Author: aqua halford dilemma If it was as simple as stealing prompts why would they need an internal model thats way smarter than the public frontier + millions in credits and thousands of agents to do it? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135038) |
|
Date: September 14th, 2026 6:59 PM Author: walnut rigor yarmulke to lie (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135897) |
|
Date: September 14th, 2026 10:55 AM Author: spectacular church building faggot firefighter (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134861) |
|
Date: September 14th, 2026 11:17 AM Author: Sticky shitlib menage “AI” didn’t solve shit, it just stole a real mathematician’s proof and that gay no-talent jew Altman claimed it as his (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134922) |
|
Date: September 14th, 2026 11:20 AM Author: aqua halford dilemma (gullible retard) (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134933) |
|
Date: September 14th, 2026 11:21 AM Author: Sticky shitlib menage Oh damn that really settles the matter lmfao (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134935) |
|
Date: September 14th, 2026 11:23 AM Author: aqua halford dilemma Durrrrr open ai released their proof so I guess the mathematician now can release his and show they are the same durrr. Except no. The exact opposite has happened and their proof is radically different from what the guy was doing other than being in the same research program. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134945) |
|
Date: September 14th, 2026 2:36 PM Author: sepia hominid Altman said the two mathematicians deserve credit while also insisting they did nothing (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135294) |
|
Date: September 14th, 2026 9:17 PM Author: Slippery spot His name? Albert Einstein (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50136194) |
|
Date: September 14th, 2026 2:16 PM Author: Cream cuckold whorehouse Nooo my tokensssss millions of tokennzsths (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135258) |
|
Date: September 14th, 2026 11:04 AM Author: sepia hominid Shoulda listened to NSAM (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134879) |
|
Date: September 14th, 2026 11:09 AM Author: sapphire juggernaut rehab 90%+ of open source projects are on Github > Microsoft > OpenAI > Pentagon (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134895) |
|
Date: September 14th, 2026 11:42 AM Author: exciting hateful location (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135000) |
|
Date: September 14th, 2026 11:25 AM Author: silver foreskin "Buckmaster and Alpöge had been using AI models to help accelerate their work. Buckmaster worked with commercially available versions of ChatGPT and Claude, paid for out of his research funds, and Alpöge worked with internal models accessible to him as an Anthropic employee."
I wanted AI to solve the problem, but not like this, not by them. |
|
Date: September 14th, 2026 11:50 AM Author: aqua halford dilemma Whats funny is that the thing they were working on isn't even the same, its just related. Alpoge and Buckmaster's work concerned forced Euler, whereas open ais result establishes blowup for unforced Navier–Stokes. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135029) |
|
Date: September 14th, 2026 12:11 PM Author: silver foreskin How does the "blow up" solve the problem? Article explained what the blow ups were and how they related to the problem, but not how they lead to solving it. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135097) |
|
Date: September 14th, 2026 12:28 PM Author: aqua halford dilemma The clay formulation is an either or and they chose route b, the part about smoothness assumptions. A globally smooth solution cannot have its velocity blow up at finite time. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135127) |
|
Date: September 14th, 2026 11:25 AM Author: supple beta becky space what’s the difference between that and AI waiting three weeks to solve it until after the academics’ work was published? It’s not like a NYU prof can burn $22 million of takes to solve it himself (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134952) |
|
Date: September 14th, 2026 11:38 AM Author: silver foreskin This is a more significant issue, if it's true: "“What I don’t think the mathematical community will accept is the write-up as it is,” Gomez-Serrano said of OpenAI’s solution, which he is currently trying to read. “It’s incomprehensible.”"
If true, OpenAI didn't solve anything. That's the story. |
|
Date: September 14th, 2026 11:40 AM Author: aqua halford dilemma You don't think open ai's internal team of mathematicians are a little bit more intelligent than the cranks trying to claim their proof was stolen? And you don't think their researchers sent it to top external mathematicians before publishing? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134994) |
|
Date: September 14th, 2026 11:41 AM Author: walnut rigor yarmulke no (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50134999) |
|
Date: September 14th, 2026 11:44 AM Author: aqua halford dilemma They do idiot. They use their internal team to review it, which includes fields medalist Jacob Tsimerman + machine proof checkers like LEAN4 and agda, and also send it externally to top mathematicians to verify. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135006) |
|
Date: September 14th, 2026 11:55 AM Author: walnut rigor yarmulke you sound really confident for a lawyer (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135050) |
|
Date: September 14th, 2026 12:07 PM Author: aqua halford dilemma They circulated the Erdos proof among their network of top external mathematicians before publishing it, and this one is much more consequential. I'm not saying its impossible that they didn't, but the alternative is pretty absurd. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135087) |
|
Date: September 14th, 2026 12:09 PM Author: walnut rigor yarmulke do u remember what year we live in? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135093) |
|
Date: September 14th, 2026 11:46 AM Author: Concupiscible fuchsia library no (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135014) |
|
Date: September 14th, 2026 12:16 PM Author: silver foreskin I'm highly skeptical of OpenAI and I don't fully trust them to have thoroughly vetted their proof rather than rushing it out to try to scoop others working on the same topic, including Anthropic. If they had, why not say it? "Professors so and so vetted the proof and confirm it's correct."
The quote above is from a professor at Brown. He may be wrong, biased, or a crank, but it's not some guy on the street. And he didn't just say he's not sure. He said it's "incomprehensible." Almost as if AI wrote it... |
|
Date: September 14th, 2026 12:25 PM Author: aqua halford dilemma The guy from Brown you are talking about is legit, but look at what he's saying. He is saying the write-up itself is hard to understand, not that he found a flaw or that he thinks its wrong. He thinks the paper wasn't well written. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135120) |
|
Date: September 14th, 2026 12:52 PM Author: silver foreskin If it's strictly about the paper and not about the proof or whether OpenAI really solved the problem, it's a pretty misleading passage in the article. It suggests there is something more going on than just a poorly drafted paper. It says the community won't accept it. And I for one would be shocked (shocked!) that the Washington Post would include a misleading and incendiary passage suggesting that there is a problem with the proof itself if that's not the case. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135158) |
|
Date: September 14th, 2026 7:33 PM Author: aqua halford dilemma Yeah the quote said the community wouldn't accept the paper itself as written, not that the mathematical content or proof was wrong. The paper is written in this overly complicated way that no normal person would ever read and understand. It isn't nonsense, but it isn't readable by humans. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135956) |
|
Date: September 14th, 2026 11:47 AM Author: spectacular church building faggot firefighter The writeup is balls but the actual proof is written in a formal mathematical language called Lean that enables full ironclad verification. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135020) |
|
Date: September 14th, 2026 12:12 PM Author: silver foreskin Does this have said full ironclad verification? If so, why is there a dispute? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135100) |
|
Date: September 14th, 2026 12:19 PM Author: aqua halford dilemma I think most people accept that it is legit. Lean isn't ironclad though. Lean guarantees that given the specific assumptions encoded from it the proof follows. It can't guarantee that the proof actually corresponds to the problem Clay was talking about in its formulation. Maybe some people will offer a legit dispute. I think its low probability that the result collapses completely. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135112) |
|
Date: September 14th, 2026 2:34 PM Author: sepia hominid Anyone got a non-AI generated take? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135291) |
|
Date: September 14th, 2026 4:56 PM Author: mind-boggling mad-dog skullcap piazza Yeah, I thought mathematicians generally accept Lean proofs once the input is verified. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135616) |
|
Date: September 14th, 2026 5:04 PM Author: aqua halford dilemma Sometimes they will argue over whether the assumptions that went into it hold (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135639) |
|
Date: September 14th, 2026 2:33 PM Author: sepia hominid ljl I've seen this happen before. SWIM thought he came up with a more elegant explanation for time dilation using Gemini, and he tried to use it to try to patent some kind of GPS device. Eventually no patent lawyer would respond to his emails (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135287) |
|
Date: September 14th, 2026 1:37 PM Author: orange crackhouse partner yes, AI is Jewish (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135210) |
|
Date: September 14th, 2026 2:13 PM Author: walnut rigor yarmulke finally someone has the balls to say it (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2\u0026mark_id=5310910#50135257) |